KR102214634B1 - Properties prediction system based on artificial intelligence - Google Patents

Abstract

The present invention relates to an artificial intelligence-based property prediction system, and in detail, by receiving material information-composition information-physical property information-use information, etc. for various composite resins, artificial intelligence is applied to various composite resin-related information stored and managed. It relates to an artificial intelligence-based physical property prediction system that can easily provide appropriate information according to user's needs by analyzing and comparing by applying an intelligence-based model generation algorithm.

Description

{Properties prediction system based on artificial intelligence}

The present invention relates to an artificial intelligence-based property prediction system, and in detail, by applying various artificial intelligence algorithms (machine learning, deep learning, etc.) based on a material-composition-physical property-use database of functional composite resin, composition and physical properties It relates to an artificial intelligence-based physical property prediction system that can predict and provide.

Composite resin refers to a material that combines two or more types of materials. In general, after compounding two or more types of materials, each material maintains its original physical/chemical image while maintaining superior performance than the original material. it means.

Therefore, it is possible to invest a lot of time and effort to predict the properties of a composite resin or to propose an optimal composition and use by complexing a number of materials.

In addition, compared to a lot of time and effort, the reality is that the accuracy or reliability of the results is low.

In this regard, Korean Patent Laid-Open No. 10-2016-0060993 ("Method and system for predicting the physical properties of a compound") describes a method that can predict the physical properties of a compound with high accuracy using a method based on the density functional theory. It is starting.

Korean Patent Publication No. 10-2016-0060993 (Publication date 2016.05.31.)

The present invention was devised to solve the above-described problems, and an object of the present invention is to apply various artificial intelligence algorithms (machine learning, deep learning, etc.) based on a material-composition-physical property-use database of a functional composite resin. It is to provide an artificial intelligence-based property prediction system that can predict and provide composition and properties.

The artificial intelligence-based physical property prediction system according to an embodiment of the present invention includes a first DB unit 10 storing material-related information on a composite resin, and stores it in the first DB unit 10 at the request of a user. A model generation unit 100 applied to an artificial intelligence-based property model generation algorithm generated based on existing information, and a property calculation unit 200 that calculates predicted property information based on the property model generated by the model generation unit 100 , It is preferable to include an output unit 300 that receives and outputs the information generated by the physical property calculation unit 200.

Furthermore, the artificial intelligence-based property prediction system further includes a second DB unit 20 that stores use-related information on the composite resin, and the predicted property among the information stored in the second DB unit 20 It is preferable to transfer the usage proposal information matched with the information to the output unit 300.

Further, the model generation unit 100 requests the user to input composition information for the composite resin, analyzes material information corresponding to the composition information according to the user's input, and sends the first DB unit 10 to It is desirable to apply it to an artificial intelligence-based physical property model generation algorithm created based on the stored information.

Furthermore, it is preferable that the artificial intelligence-based physical property prediction system further includes a composition generator 400 that limits the composition by using composition-related information on the composite resin.

Furthermore, the model generation unit 100 requests the user to input composition range information or physical property information for the composite resin, and according to the user's input, the composition generation unit 400 generates the composition range information or property information. It is preferable to receive, limit the composition, analyze material information corresponding to the limited composition information, and apply it to an artificial intelligence-based physical property model generation algorithm generated based on the information stored in the first DB unit 10.

Furthermore, the artificial intelligence-based property prediction system further includes a property comparison and determination unit 500 for comparing and determining predicted property information by the property model calculated by the property calculation unit 200, and the property comparison and determination unit It is preferable to generate optimal composition information according to the determination result of 400 and transmit it to the output unit 300.

Furthermore, the model generation unit 100 requests the user to input use information for the composite resin, and according to the user's input, the model generation unit 100 matches the use information among the information stored in the second DB unit 20. By extracting the use specification information, analyzing the requested physical property information for each of the extracted use information, limiting the composition by receiving the requested property information from the composition generating unit 400, and analyzing the material information corresponding to the limited composition information , Applying to an artificial intelligence-based physical property model generation algorithm generated based on the information stored in the first DB unit 10, and comparing and determining the predicted property information by the property model in the property comparison and determination unit 400, It is preferable to generate optimal composition information according to the determination result and transmit it to the output unit 300.

Furthermore, the artificial intelligence-based property prediction system further includes a third DB unit 30 that stores environmental regulation-related information on the composite resin, and is input among information stored in the third DB unit 30. It is preferable to transmit composition information or environmental regulation information matched with the generated optimal composition information to the output unit 300.

Furthermore, the artificial intelligence-based physical property prediction system further includes a fourth DB unit 40 that stores price-related information on the composite resin, and the input composition among the information stored in the fourth DB unit 40 It is preferable to transmit information or price information matched with the generated optimal composition information to the output unit 300.

Furthermore, the artificial intelligence-based physical property model generation algorithm of the model generation unit 100 includes multiple linear regression (MLR), support vector machine (SVM), and nearest neighbor classification (k-NN). , k-Nearest Neighbor), Deep Learning, Genetic Algorithm (GA), Boosted Trees, Generative Adversarial Network (GAN), Artificial Neural Network (ANN) ) It is preferable that it is any one or more selected from.

The artificial intelligence-based physical property prediction system of the present invention can be provided by predicting composition and physical properties by applying various artificial intelligence algorithms (machine learning, deep learning, etc.) based on the material-composition-physical property-use database of functional composite resin. There is an advantage.

In detail, in order to make a composition-physical property prediction model, various artificial intelligence algorithms can be applied, and when the artificial intelligence algorithm is applied, characteristic engineering techniques can be applied to improve the accuracy of prediction.

In other words, it is possible to save the time and effort required for predicting physical properties or extracting the optimal composition by making the main characteristics in various ways (applied to various artificial intelligence algorithms) and analyzing the effect of the characteristics on the physical properties and using them for the development of important characteristics. have.

Particularly, a part of the material-composition-physical property-use database information may be used as a test set, thereby verifying the accuracy of the generated prediction model.

Through this, it is possible to derive major characteristics by analyzing various characteristics of various composite resin information received from the user through feature engineering, and applying this to an artificial intelligence-based physical property model generation algorithm, predicting property information. /As it can provide optimal composition information, it is possible to save time and effort required for predicting properties or extracting the optimum composition.

Furthermore, environmental regulation information (using raw materials/mixtures) and price information can be provided, so it can be used as an integrated prediction system based on the material-composition-physical property-use database of composite resins.

1 is a schematic diagram of an artificial intelligence-based physical property prediction system according to an embodiment of the present invention.
2 to 5 are diagrams showing various embodiments of an artificial intelligence-based physical property prediction system according to an embodiment of the present invention.

Hereinafter, an artificial intelligence-based physical property prediction system according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings. The drawings introduced below are provided as examples in order to sufficiently convey the spirit of the present invention to those skilled in the art. Accordingly, the present invention is not limited to the drawings presented below and may be embodied in other forms. In addition, the same reference numbers throughout the specification indicate the same elements.

In this case, unless there are other definitions in the technical terms and scientific terms used, they have the meanings commonly understood by those of ordinary skill in the art to which this invention belongs, and the gist of the present invention in the following description and accompanying drawings A description of known functions and configurations that may unnecessarily obscure will be omitted.

In addition, the system refers to a set of components including devices, devices, and means that are organized and regularly interact to perform a required function.

The artificial intelligence-based property prediction system according to an embodiment of the present invention receives material information-composition information-physical property information-use information, etc. for various composite resins, and analyzes and stores various composite resin-related information. By comparing, it is an artificial intelligence-based physical property prediction system that can easily provide appropriate information according to the user's request.

FIG. 1 is a block diagram showing an artificial intelligence-based property prediction system according to an embodiment of the present invention, and an artificial intelligence-based property prediction system according to an embodiment of the present invention will be described in detail with reference to FIG. 1.

As shown in FIG. 1, the artificial intelligence-based property prediction system according to an embodiment of the present invention includes a model generation unit 100, a property calculation unit 200, an output unit 300, a composition generation unit 400, and a property. It is preferable to include a comparison and determination unit 500, and the output unit 300 refers to a terminal capable of receiving information on various composite resins, such as a terminal possessed by a user and a terminal possessed by an administrator. And, the output unit 300, the model generation unit 100, and the physical property calculation unit 200 may perform communication using a wireless or wired network to input a user request in real time or receive judgment information accordingly. have.

In addition, it is preferable that the model generation unit 100 and the physical property calculation unit 200 are configured and applied to one main controller unit (MCU).

At this time, in order to perform communication, the model generation unit 100, the property calculation unit 200, the output unit 300, the composition generation unit 400, and the property comparison and determination unit 500 are each a network management unit (not shown). City), and when performing communication using a wireless network, the network management unit may include a wireless LAN (WLAN, Wireless LAN), Wi-Fi (Wireless Fidelity), and a wireless broadband Internet. ), WiMAX (World Interoperability for Microwave Access), HSDPA (High Speed Downlink Packet Access), Zigbee, Bluetooth, UWB (Ultra-WideBand), IrDA (Infrared Data Association), Ultra Wild Band ), SWAP (Shared Wireless Access Protocol), LTE (Long Term Evolution), 5G (Fifth-Generation), etc. to perform wireless communication.

Prior to learning about each configuration in detail, the artificial intelligence-based physical property prediction system according to an embodiment of the present invention includes a first DB unit 10 storing material-related information on a composite resin, as shown in FIG. 1. It is preferable to be configured to include.

It is preferable that the first DB unit 10 is configured and applied to one MCU together with the model generation unit 100 and the physical property calculation unit 200.

It is preferable that the model generation unit 100 applies the user's input information to an artificial intelligence-based physical property model algorithm generated based on the information stored in the first DB unit 10.

It is preferable that the physical property calculation unit 200 calculates predicted property information by the property model generated by the model generation unit 100, and transmits and outputs the information generated by the output unit 300. desirable.

Here, as the artificial intelligence-based physical property model generation algorithm stored in advance in the model generation unit 100, multiple linear regression (MLR), support vector machine (SVM), and nearest neighbors Classification (k-NN, k-nearest neighbor), Deep learning, Genetic Algorithm (GA), Boosted Trees, Generative Adversarial Network (GAN), Artificial Neural Network ( ANN, Artificial Neural Network), ensemble, etc., and it is preferable to generate a physical property model by applying any one or more selected from the above algorithms.

Specifically, the model generation unit 100 comprises a matrix (mainly polymer), a dispersed phase (mainly polymer, filler), additives, and other components, contents (mainly volume fraction, mass fraction), which are materials (components) of plastic composite resin. , It is desirable to predict composition or physical properties by applying features such as intrinsic properties and process conditions to generate a new property model or automatically learn feature extraction through learning.

In the case of the multiple linear regression analysis, a method of predicting output properties (predicted properties) by calculating a correlation coefficient by applying input properties (material, content, process conditions, etc.) to a multiple plastic regression analysis model created through training data. Conversely, it is possible to predict the composition through the model by putting physical properties as input properties.

In the case of the support vector machine, a hyperplane capable of classifying data in a three-dimensional space is created by applying an input attribute to the kernel function of the support vector machine generated through the training data, and then data close to the hyperplane. (Support Vector) This is a method of predicting proximity properties by calculating the margin of the feature, and types of kernel functions include a linear kernel, a polynomial kernel, a Gaussian kernel, an inverse tangent kernel, and a sigmoid kernel.

In the case of the nearest neighbor classification, the proximity measure is calculated by applying the input attribute to the learned nearest neighbor classification model, and the predicted property of the input attribute is found by determining the nearest classification (material, content, process conditions, etc.). Says the algorithm. That is, after calculating the distance similarity between data, the predicted properties can be found by determining a number of nearest neighbors among k neighbors. On the contrary, it is also possible to find out the predicted introduction and content by putting physical properties as input properties.

In the case of the deep learning, feature extracts are automatically extracted from the hidden layer generated by applying the input attributes to the learned deep learning model to calculate the feature strength according to the input attributes, and this is reflected in the active function of the hidden layer. Thus, the physical properties, which are output properties, can be predicted. The activation functions are LogSigmoid, SeLU, ReLu, Softplus, Logsoftplusmax, Softshrink, Hyperbolic tanhshrink. ), Softsign-Y, Logistic, Softmax, etc. are applicable, and optimization algorithms include Adagrad, Adam, Adamax, RMSProp, Rprop, SparseAdam, ASGD, SGD, LBFGS, etc. It is possible. It is also possible to find out the predicted composition for the input property properties as a result.

In the case of the genetic algorithm, a variety of functional expressions are created by combining and applying several input attributes to the learned genetic algorithm model, and after evolutionary calculations (selection, intersection, mutation, substitution, etc.) are performed, the main variables are extracted to predict optimal properties. Explore the model. In addition, on the contrary, the predicted composition for the input property properties can be found as a result.

In the case of the boost tree, a more accurate and powerful learner is created by combining simple and weak learners in machine learning in which input attributes are introduced into the learned boost tree model. Even if the accuracy is low, once a model is created and the weaknesses revealed are compensated through the second model, when the two are combined, an accurate model is created rather than the first, and the problem that remains is the principle of repeating the process of supplementing and adding subordination in the next model. .

In the case of the generative adversarial neural network, it refers to a model that generates a generator that generates virtual material data by applying input properties to the learned generative adversarial neural network model and a discriminator that distinguishes it from real material data. , The generator tries to minimize the accuracy of the classifier, and the classifier can predict the output property according to the input property by competing with each other to maximize their accuracy. On the contrary, it is possible to find out the predicted composition for the input property properties as a result.

In the case of the artificial neural network, the synapse (feature) that learns the relationship between the input layer and the output layer by applying the input attribute to the learned artificial neural network creates artificial neurons with different binding strengths, and through the learned artificial neurons. Physical properties (or composition) according to input properties can be predicted.

In the case of the above ensemble, it refers to a method of generating various predictive models by applying input attributes to the learned ensemble, and combining the features generated from the models. Correlation coefficient voting on the features of predictive models for new input attributes To create a new model. By combining the prediction results of various models, more reliable predictive properties can be obtained when analyzed with a single model.

As shown in FIG. 1, the system for predicting physical properties based on artificial intelligence according to an embodiment of the present invention is preferably configured to further include a second DB unit 20 that stores information related to use of the composite resin.

Like the first DB unit 10, the second DB unit 20 is preferably applied to a single MCU together with the model generation unit 100 and the physical property calculation unit 200.

It is preferable that the physical property calculation unit 200 transmits, to the output unit 300, usage proposal information that is matched with the predicted property information among information stored in the second DB unit 20.

Embodiment 1 (entering composition information)

As shown in FIG. 2, the artificial intelligence-based property prediction system according to an embodiment of the present invention may output predicted property information and usage proposal information most optimized for input composition information according to a user's request.

Specifically, as shown in FIG. 2, it is preferable to analyze material information corresponding to the composition information as the user inputs composition information on the composite resin.

The model generation unit 100 may be applied to an artificial intelligence-based physical property model generation algorithm generated based on the information stored in the first DB unit 10 to generate a physical property model.

After that, it is preferable that the physical property calculation unit 200 calculates the predicted property information by the property model generated by the model generation unit 100 and transmits it to the output unit 300 for output.

At this time, in order to increase accuracy and reliability, it is possible to request the user to input not only composition information on the composite resin, but also a limiting factor.

In other words, it is possible to request an input of a property limiting element to limit it to a specific property, and to request an input of a process limiting factor to limit it to a specific process. In addition, it is possible to request input of weight setting. Weight setting means that when a composition + physical property is combined, an input can be requested to present the physical property (or composition) by extracting a similar blend according to the weight (for example, a filler in composition or tensile strength among properties). have.

In addition, the composition information that can be inputted from the user includes 1. polymer or filler, additive, 2. 1. content, weight ratio or volume ratio, 3. 1. shape, structure, size or length, 4. 1. orientation , 5. Material properties, density/specific gravity, surface energy, surface area, mechanical properties, electrical properties, thermal properties, etc. can be input.

Process limiting factors that can be input from the user include: 1. temperature, 2. pressure, 3. screw combination/speed, 4. equipment type, 5. resin input amount, 6. inlet location, 7. residence time ((residence time) ) = f(screw speed(rpm), screw configuration, feeding rate)), etc. can be input,

As the predicted property information that can be calculated by the property calculation unit 200 by receiving such information, 1.mechanical properties (tensile strength, elongation, flexural strength, flexural modulus, impact strength, surface strength, etc.), 2. Physical properties (density, melting index, etc.), 3. Thermal properties (heat deformation temperature, thermal conductivity, glass transition temperature, etc.), 4. Electrical properties (electric conductivity, dielectric constant, etc.), 5. Multi-point graph data (stress- Deformation, modulus of elasticity, etc.), and image properties (SEM, TEM, etc.) may be exemplified, but these are only an embodiment of the present invention.

Second embodiment (enter composition range)

It is preferable that the artificial intelligence-based physical property prediction system according to an embodiment of the present invention further includes a composition generating unit 400 as shown in FIG. 3.

The composition generator 400 may limit the composition by using composition-related information on the composite resin. In other words, the composition generator 400 is a composition generator and may limit a composition.

Including the composition generating unit 400, as shown in FIG. 3, according to the user's request, it is possible to output the most optimized predicted physical property information and use suggestion information for the input composition range information.

In detail, as shown in FIG. 3, it is preferable to limit the composition in the composition generating unit 400 according to the user's input information on the composite resin (composition range information input).

Material information corresponding to the limited composition information is preferably applied to the model generation unit 100 to which the artificial intelligence-based physical property model generation algorithm generated from the first DB unit 10 is applied.

After that, it is preferable that the physical property calculation unit 200 calculates the predicted property information by the property model generated by the model generation unit 100 and transmits it to the output unit 300 for output.

At this time, in order to increase accuracy and reliability, as in the first embodiment described above, it is possible to request the user to input not only the composition range information on the composite resin, but also the limiting factor.

In other words, it is possible to request an input of a property limiting element to limit it to a specific property, and to request an input of a process limiting factor to limit it to a specific process. In addition, it is possible to request input of weight setting. Weight setting means that when a composition + physical property is combined, an input can be requested to present the physical property (or composition) by extracting a similar blend according to the weight (for example, a filler in composition or tensile strength among properties). have.

Third embodiment (input of property information)

As shown in FIG. 4, the artificial intelligence-based property prediction system according to an embodiment of the present invention may output composition and usage proposal information most optimized for input property information according to a user's request.

In addition, as shown in FIG. 4, it is preferable to further include a physical property comparison and determination unit 500.

The property comparison and determination unit 500 compares and determines the predicted property information by the property model calculated by the property calculation unit 200, generates optimal composition information according to the determination result, and transmits it to the output unit 300. It is desirable to do.

Specifically, as shown in FIG. 4, after requesting the user to input physical property information on the composite resin and receiving the user's input (physical property information input), it is preferable to limit the composition in the composition generating unit 400 .

It is preferable to apply material information corresponding to the limited composition information to the model generation unit 100 to which an artificial intelligence-based physical property model generation algorithm generated from the first DB unit 10 is applied.

After that, it is preferable that the physical property calculation unit 200 calculates the predicted property information by the property model generated by the model generation unit 100 and transmits it to the output unit 300 for output. In this case, the predicted material information is the limited predicted property information for each composition information.

In addition, the physical property comparison and determination unit 500 compares and determines the predicted property information by the property model calculated by the property calculation unit 200, and generates optimal composition information according to the determination result, and the output unit 300 ).

In addition, in order to increase accuracy and reliability, as in the first embodiment described above, it is possible to request the user to input not only physical property information on the composite resin, but also a limiting factor. That is, the input of the composition limiting element may be requested to be limited to a specific composition, and the input of the process limiting factor may be requested to be limited to a specific process. In addition, it is possible to request input of weight setting. Weight setting means that when a composition + physical property is combined, an input can be requested to present the physical property (or composition) by extracting a similar blend according to the weight (for example, a filler in composition or tensile strength among properties). have.

Fourth embodiment (application information input)

As shown in FIG. 5, the artificial intelligence-based property prediction system according to an embodiment of the present invention may output predicted property information and usage proposal information most optimized for input usage information according to a user's request.

In detail, as shown in FIG. 5, after requesting the user to input use information for the composite resin and receiving the user's input (application information input), it is preferable to limit the composition in the composition generating unit 400 .

In addition, the model generation unit 100 extracts usage specification information matched with the usage information from among the information stored in the second DB unit 20, and analyzes the required physical property information required for each of the extracted usage information. I can.

It is preferable to apply material information corresponding to the limited composition information to the model generation unit 100 to which an artificial intelligence-based physical property model generation algorithm generated from the first DB unit 10 is applied.

After that, it is preferable that the physical property calculation unit 200 calculates the predicted property information by the property model generated by the model generation unit 100 and transmits it to the output unit 300 for output. In this case, the predicted material information is the limited predicted property information for each composition information.

In addition, the physical property comparison and determination unit 500 compares and determines the predicted property information by the property model calculated by the property calculation unit 200, and generates optimal composition information according to the determination result, and the output unit 300 ).

In addition, in order to increase accuracy and reliability, it is possible to request the user to input not only usage information for the composite resin, but also a limiting factor. That is, it is possible to request an input of a property limiting element to limit it to a specific property, to request an input of a composition limiting factor to limit it to a specific composition, and to request an input of a process limiting factor to limit it to a specific process. In addition, it is possible to request input of weight setting. Weight setting means that when a composition + physical property is combined, an input can be requested to present the physical property (or composition) by extracting a similar blend according to the weight (for example, a filler in composition or tensile strength among properties). have.

In addition, the AI-based physical property prediction system according to an embodiment of the present invention may further include a third DB unit 30 and a fourth DB unit 40 as shown in FIG. 1.

The third DB unit 30 preferably stores environmental regulation-related information on the composite resin, and the fourth DB unit 40 preferably stores price-related information on the composite resin.

It is preferable that the third DB unit 30 and the fourth DB unit 40 are also configured and applied to one MCU together with the model generation unit 100 and the physical property calculation unit 200.

In the artificial intelligence-based physical property prediction system according to an embodiment of the present invention, the third DB unit 30 and the fourth DB unit 40 are stored in the third DB unit 30 under the control of the configured MCU. Among the information, the input composition information or environmental regulation information matching with the generated optimal composition information may be extracted and transmitted to the output unit 300.

In addition, among the information stored in the fourth DB unit 40, input composition information or price information matching the generated optimal composition information may be extracted and transmitted to the output unit 300.

That is, in other words, the artificial intelligence-based property prediction system according to an embodiment of the present invention receives material information-composition information-physical property information-use information for various composite resins, and stores and manages various composite resins. By analyzing / comparing / matching related information, it is possible to easily provide appropriate information according to the user's request.

Meanwhile, the artificial intelligence-based property prediction system according to an embodiment of the present invention may be implemented in the form of a program command that can be executed through various electronic information processing means and recorded in a storage medium. The storage medium may include a program command, a data file, a data structure, or the like alone or in combination.

The program instructions recorded in the storage medium may be specially designed and configured for the present invention, or may be known and usable to those skilled in the software field. Examples of storage media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic-optical media such as floptical disks. hardware devices specially configured to store and execute program instructions such as magneto-optical media and ROM, RAM, flash memory, and the like. Further, the above-described medium may be a transmission medium such as an optical or metal wire, a waveguide including a carrier wave for transmitting a signal specifying a program command, a data structure, or the like. Examples of program instructions include not only machine language codes such as those produced by a compiler, but also a device that processes information electronically using an interpreter, for example, a high-level language code that can be executed by a computer.

Although the above has been described with reference to preferred embodiments of the present invention, those of ordinary skill in the relevant technical field can variously modify the present invention within the scope not departing from the spirit and scope of the present invention described in the following claims. It will be appreciated that it can be modified and changed.

As described above, in the present invention, specific matters such as specific components and the like have been described by the drawings of limited embodiments, but this is provided only to help a more general understanding of the present invention, and the present invention is limited to the above-described embodiment. It is not, and those of ordinary skill in the field to which the present invention belongs can make various modifications and variations from this description.

Therefore, the spirit of the present invention is limited to the described embodiments and should not be determined, and all things having equivalent or equivalent modifications to the claims as well as the claims to be described later belong to the scope of the spirit of the present invention. will be.

10: 1st DB part 20: 2nd DB part
30: 3rd DB part 40: 4th DB part
100: model generation unit
200: Physical property calculation unit
300: output
400: composition generating unit
500: physical property comparison and judgment part

Claims (10)

A first DB unit 10 storing material-related information on the composite resin;
A model generation unit 100 for generating a physical property model to which a pre-stored artificial intelligence-based physical property model generation algorithm is applied based on information stored in the first DB unit 10 according to a user's request;
A property calculation unit 200 for calculating predicted property information based on the property model generated by the model generation unit 100;
An output unit 300 for receiving and outputting the information generated by the physical property calculation unit 200;
A composition generating unit 400 for limiting the composition by using composition-related information on the composite resin; And
A property comparison and determination unit 500 for comparing and determining predicted property information based on the property model calculated by the property calculation unit 200;
Consists of including,
The physical property calculation unit 200 requests the user to input physical property information on the composite resin and transmits the received physical property information to the composition generation unit 400,
The composition generating unit 400 limits a composition for the received physical property information,
The physical property calculation unit 200 applies material information corresponding to the limited composition information to the physical property model, calculates the predicted property information for each composition, and transmits it to the output unit 300,
The property comparison and determination unit 500 compares the predicted property information for each composition calculated, determines optimal composition information based on the received property information, and transmits the determination result to the output unit 300. Artificial intelligence-based property prediction system.
The method of claim 1,
The artificial intelligence-based property prediction system
A second DB unit 20 storing information related to use of the composite resin;
It further includes,
An artificial intelligence-based material property prediction system, characterized in that, among the information stored in the second DB unit (20), the use proposal information matched with the predicted property information is transmitted to the output unit (300).
The method of claim 2,
The model generation unit 100
Request the user to input composition information for the composite resin, and according to the user's input,
An artificial intelligence-based physical property prediction system, characterized in that analyzing material information corresponding to the composition information and applying it to an artificial intelligence-based physical property model generation algorithm generated based on the information stored in the first DB unit (10).
delete The method of claim 2,
The model generation unit 100
Request the user to input composition range information or physical property information for the composite resin, and according to the user's input,
By receiving the composition range information or physical property information from the composition generating unit 400, the composition is limited,
An artificial intelligence-based physical property prediction system, characterized in that the material information corresponding to the limited composition information is analyzed and applied to an artificial intelligence-based physical property model generation algorithm generated based on the information stored in the first DB unit 10.
delete delete The method according to any one of claims 1 to 3 and 5,
The artificial intelligence-based property prediction system
A third DB unit 30 that stores environmental regulation-related information on the composite resin;
It further includes,
An artificial intelligence-based physical property prediction system, characterized in that, among the information stored in the third DB unit 30, the input composition information or environmental regulation information matched with the generated optimal composition information is transmitted to the output unit 300. .
The method according to any one of claims 1 to 3 and 5,
The artificial intelligence-based property prediction system
A fourth DB unit 40 storing price-related information on the composite resin;
It further includes,
An artificial intelligence-based physical property prediction system, characterized in that, among information stored in the fourth DB unit (40), the input composition information or price information matched with the generated optimum composition information is transmitted to the output unit (300).
The method of claim 1,
The artificial intelligence-based physical property model generation algorithm of the model generation unit 100
Multiple Linear Regression (MLR), Support Vector Machine (SVM), Nearest Neighbor Classification (k-NN, k-Nearest Neighbor), Deep Learning, Genetic Algorithm (GA, Generic Algorithm), Boosted Trees, Generative Adversarial Network (GAN), and Artificial Neural Network (ANN).

Images